How can I create groups based on ascending streak of a column?

This is my DataFrame:

import pandas as pd
df = pd.DataFrame(
    {
        'a': [10, 14, 20, 10, 12, 5, 3]
    }
)

And this is the expected output. I want to create four groups:

    a
0  10
1  14
2  20

    a
3  10
4  12

    a
5   5

    a 
6   3 

From top to bottom, as far as a is increasing/equal groups do not change. That is why first three rows are in one group. However, in row 3 a has decreased (i.e. 20 > 10). So this is where the second group starts. And the same logic applies for the rest of groups.

This is one my attempts. But I don’t know how to continue:

import numpy as np
df['dir'] = np.sign(df.a.shift(-1) - df.a)

>Solution :

Code

maybe there are 4 groups

dfs = [d for _, d in df.groupby(df['a'].diff().lt(0).cumsum())]

dfs

[    a
 0  10
 1  14
 2  20,
     a
 3  10
 4  12,
    a
 5  5,
    a
 6  3]